Recurrent Tonsillitis in Adults

Abstract

Background

The aim of this study was to assess the effect of tonsillectomy in adults with recurrent tonsillitis on their quality of life and on their use of medical resources.

Method

114 patients who had had at least three episodes of acute tonsillitis in the 12 months preceding tonsillectomy were evaluated pre- and postoperatively with a questionnaire developed by the authors, and with the Glasgow Benefit Inventory.

Results

97 patients (85%) filled out the questionnaires completely. The Glasgow Benefit Inventory revealed an improvement in the overall score (+19) and in the partial scores for general well-being (+18) and physical health (+39). The degree of support from friends and family was unchanged (±0). Significant decreases were observed in visits to a physician, analgesic and antibiotic consumption, days off from work, and episodes of sore throat. The number of visits to a physician because of sore throat decreased from an average of five preoperatively to one postoperatively; the number of episodes of sore throat, from seven to two; and the number of days taken off from work, from twelve to one per year. 65% of the patients surveyed took analgesics for sore throat preoperatively, 7% postoperatively. 95% took antibiotics for sore throat preoperatively, 22% postoperatively.

Conclusion

Although this study had a number of limitations (small size, retrospective design, short follow-up), it was able to show that tonsillectomy for adults with recurrent tonsillitis improves health and quality of life and reduces the need to consume medical resources.

Tonsillectomy is one of the most frequently performed surgical operations in Germany (1). While many investigations of various techniques, postsurgical bleeding rates, and postoperative pain have been published, few studies throw any light on the outcome of surgery with regard to the reason the operation was performed, namely painful throat infections. Of particular interest are the effect of tonsillectomy on the number of visits to a doctor, the necessity and frequency of analgesic and antibiotic treatment, and the number of days off work because of illness.

Although the need for well-substantiated data is self-evident, few studies to date have concerned themselves with this topic. In January 2008, a Medline search using the terms “life quality AND tonsillectomy”, “benefit AND tonsillectomy”, and “economic AND tonsillectomy” revealed only 11 relevant publications. On closer inspection, many of these studies looked mainly at the effectiveness of the procedure with regard to the objective clinical postoperative outcome (2– 5). The patients’ subjective impression of the outcome of tonsillectomy is often not investigated. Four of the 11 studies were restricted to children, and only one (6) prospectively investigated the most frequent indication for tonsillectomy: chronic or recurrent tonsillitis. In this respect, the cost efficiency and the objective clinical postoperative improvement in symptoms after tonsillectomy have been demonstrated (3, 4, 7). However, the patient’s opinion regarding the success of a given treatment should be ascribed equal importance to the clinician’s assessment (8). In the present study we set out to look at the patient’s verdict and quality of life in connection with the clinical success of tonsillectomy. Our intention was to make a contribution to patient-oriented research and elucidate the health economic aspects of this operation.

Methods

A total of 114 consecutive patients were questioned preoperatively, of whom 97 could be traced and interviewed postoperatively. All of these patients underwent elective tonsillectomy at the Marienhospital Gelsenkirchen during the 11-month study period in 2004. The inclusion criteria were adult age, good knowledge of German, and at least three episodes of acute tonsillitis in the previous 12 months. The exclusion criteria were abscess tonsillectomy, tonsillectomy for suspected or proven neoplasia, exclusive hyperplasia of the tonsils, and tonsillectomy à chaud.

• On the day before tonsillectomy, after being informed about the planned procedure and giving their written consent, the patients were questioned in a structured interview. We used a questionnaire of our own design, which, in contrast to the Glasgow Benefit Inventory (GBI), permits controlled before–after comparison. This questionnaire comprises 11 questions, covering the frequency of episodes of sore throat, the frequency of visits to a physician because of these episodes, consumption of antibiotics and analgesics, and number of days of work incapacity. We found no comparable, validated questionnaire in the literature. The same questions were put to the patients 14 months after tonsillectomy (2004–2005) in a standardized telephone interview in which the period assessed was the previous 12 months, in order to exclude symptoms experienced during the postoperative healing phase. After three unsuccessful attempts to contact a patient by telephone, a letter was sent. At this second time point the patients were also surveyed by means of the GBI, a tool that was developed in 1996 to evaluate the benefit of otorhinolaryngological interventions and can be used for various procedures (9, 10). The GBI charts changes in health status and quality of life after a surgical intervention and has been validated for this purpose (11). Each of the 18 questions is answered using a five-point Likert scale, ranging from distinct improvement to distinct worsening. In addition to the total score there are three subscales, as follows:

  • Attitude to life in general (“general:” 12 questions)

  • Relationships with family and friends (“social support:” 3 questions)

  • “Physical health:” (3 questions).

For all four scales the possible scores range from –100 to +100. (These are derived as follows: first the individual Likert scores [from 1 to 5] are added up, then the total is divided by the number of questions answered, 3 is subtracted from the result, and the number yielded is multiplied by 50). Positive values represent an improvement in quality of life after the intervention, negative values a deterioration.

The tonsillectomies were performed by surgeons with varying degrees of qualification. All operations were carried out by the cold steel method with the patient under general anesthesia (12, 13). In this technique relieving incisions of the mucosa are made with scissors and then a raspatory is used to remove the tonsil from the tonsillar fossa.

We are currently conducting a study of the principal complications and adverse effects of tonsillectomy (bleeding and pain) (14). Our impression is that they are in the normal range.

Statistical methods

The patients’ characteristics were drawn up in text and frequency tables and, where necessary, compared by means of simple univariate tests (Wilcoxon rank sum test and chi-square test). The level of significance was set at = 0.05 before commencement of analysis. All resulting p values were given descriptively without further alpha adjustment. Differences in pre-and postoperative status (prescription of analgesics and antibiotics, episodes of sore throat, visits to a doctor, and days off work) were analyzed using a two-sided paired Wilcoxon rank sum test.

This study was approved by the ethics committee of the Faculty of Medicine, University of Duisburg–Essen.

Results

One hundred fourteen patients who underwent tonsillectomy in 2004 met the inclusion criteria and were surveyed. A further 467 patients who had their tonsils removed in the same period had to be excluded because, for example, tonsillectomy was performed to deal with an abscess or tumor, they were not yet 18 years old, they were not fluent German speakers, or they did not consent to participate in the study (see inclusion criteria above). Of these 114 patients, 97 (85%) could be interviewed 14 months later and their responses analyzed. These 97 patients comprised 73 (75%) women and 24 (25%) men. The sex distribution of adult tonsillectomized patients was 41% men and 59% women. The youngest patient was 18, the oldest 62 years of age. The median age was 26 years (mean 28 years, range 44 years). The average age did not differ significantly between men (28.2 years) and women (27.9 years).

Before operation the median number of attacks of tonsillitis or episodes of sore throat was six per year (mean: 7); following tonsillectomy it fell to one (mean: 2). Table 1 shows detailed data. The difference between preoperative and postoperative frequency of sore throat was highly significant (p = 0.0001).

Table 1. Sore throat episodes (per year).

		Preoperative			Postoperative
		3–6	7–10	>10	None	1–3	4–6	7–10	>10
Sex
Male	n	19	3	2	12	8	1	3	0
	Percent	79	13	8	50	33	4	13	0
Female	n	36	27	10	25	33	9	3	3
	Percent	50	37	13	34	45	12	4	4
Total	n	55	30	12	37	41	10	6	3	p<0.0001
Chi-square test for sex distribution		Test statistic		6.9	Test statistic				5.99
		Degrees			Degrees
		of freedom		2	of freedom				4
		p value		0.075	p value				0.2002

Preoperatively, 51 (53%) of those surveyed needed analgesic medication (table 2). On average, these patients had taken an analgesic 1.6 times in the previous year. After tonsillectomy only seven (7%) patients required such medication. The decrease in consumption was significant (p<0.001).

Table 2. Number of patients who took pain medication (per year).

		Preoperative		Postoperative
		No	Yes	No	Yes
Sex
Male	n	13	11	23	1
	Percent	54	46	96	4
Female	n	33	40	67	6
	Percent	45	55	92	8
Total	n	46	51	90	7	p<0.001
Chi-square test for sex distribution		Test statistic	0.2778	Test statistic	0.0445
		Degrees		Degrees
		of freedom	1	of freedom	1
		p value	0.5981	p value	0.833

The consumption of antibiotics as treatment for sore throat also decreased following tonsillectomy (table 3). Ninety-three (96%) patients took antibiotic medication before operation, only 21 (22%) thereafter. The difference was significant (p<0.01).

Table 3. Number of patients who took antibiotics (per year).

		Preoperative		Postoperative
		No	Yes	No	Yes
Sex
Male	n	1	23	21	3
	Percent	4	96	88	13
Female	n	3	70	55	18
	Percent	4	96	75	25
Total	n	4	93	76	21	p<0.01
Chi-square test for sex distribution		Test statistic	0.3358	Test statistic	0.9387
		Degrees		Degrees
		of freedom	1	of freedom	1
		p value	0.5623	P value	0.3326

Before tonsillectomy the median number of visits to a physician was four (mean: 5); thereafter, 0 (mean: 0.7). Following the operation 71 (73%) of the patients no longer needed to consult their family doctor or an ENT specialist because of sore throat. Details can be found in Table 4. The postoperative decrease in visits to a doctor was significant (p<0.001).

Table 4. Visits to a doctor (per year).

		Preoperative				Postoperative
		1–3	4–6	7–10	> 10	1–3	4–6	7–10	> 10	None
Sex
Male	n	10	11	3	0	4	0	0	0	20
	Percent	42	46	12	0	16	0	0	0	84
Female	n	22	31	17	3	15	5	2	0	51
	Percent	30	43	23	4	21	7	3	0	69
Total	n	32	42	20	3	19	5	2	0	71	p<0.001
Chi-square test for sex distribution		Test statistic			3.3222	Test statistic				3
		Degrees of freedom			3	Degrees of freedom				3
		p value			0.5054	p value				0.4092

The median number of days off work because of tonsillitis or sore throat fell from 10 days (mean: 12) before tonsillectomy to 0 days (mean: 1) postoperatively. The difference was significant (p<0.001). Details can be found in Table 5.

Table 5. Days off work (per year).

		Preoperative					Postoperative
		1–3	4–6	7–10	> 10	None	1–3	4–6	7–10	> 10	None
Sex
Male	n	5	1	4	11	3	1	0	1	0	22
	Percent	21	4	17	46	13	4	0	4	0	92
Female	n	4	6	10	32	21	1	2	4	5	61
	Percent	5	8	14	44	29	1	3	5	7	84
Total	n	9	7	14	43	24	2	2	5	5	83	p<0.001
Chi-square test for sex distribution		Test statistic				7.0584	Test statistic				3.1856
		Degrees of freedom				4	Degrees of freedom				4
		p value				0.1328	p value				0.5273

The mean total GBI score for all patients after operation was +19. The mean values for the three subscales were as follows: “general” (12 questions): +18; “physical health” (three questions): +39; “social support,” i.e., support by family and friends (three questions): +0.17. With regard to this last subscale, only seven patients reported any change from the situation before tonsillectomy (Figures 1 and 2, Table 6).

Figure 1.

The results of the GBI survey of all patients. The graph shows the total score and the scores for the three subscales: general attitude to life (General), improvement in physical health (Physical), and relationships with friends and family (Social). Positive values indicate an improvement in quality of life, negative values a worsening

Figure2.

The results of the GBI survey of all patients. The graph shows the total scores and the scores for the three subscales: general attitude to life (General), improvement in physical health (Physical), and relationships with friends and family (Social). Positive values indicate an improvement in quality of life, negative values a worsening

Table 6. Glasgow Benefit Inventory scores.

		Total score	General	Social	Physical
Total	Minimum	–25	–29.17	–66.67	–50
	Lower quartile	11.11	8.33	0	33.33
	Median	16.67	16.67	0	33.33
	Arithmetic mean	18.52	18.1	0.17	38.54
	Upper quartile	25	25	0	50
	Maximum	55.56	70.83	50	83.33
Men	Minimum	–22.22	–29.17	–16.67	33.33
	Lower quartile	11.11	8.33	0	33.33
	Median	18.06	16.67	0	33.33
	Arithmetic mean	16.44	15.8	0	35.42
	Upper quartile	22.92	25	0	50
	Maximum	33.33	45.83	16.67	66.67
Women	Minimum	–25	–25	–66.67	–50
	Lower quartile	11.11	8.33	0	33.33
	Median	16.67	16.67	0	33.33
	Arithmetic mean	19.21	18.87	0.23	39.58
	Upper quartile	25	26.04	0	54.17
	Maximum	55.56	70.83	50	83.33

Discussion

Tonsillectomy is one of the most common surgical operations in Germany, performed around 115 000 times each year (1). The diagnoses most frequently leading to tonsillectomy are chronic tonsillitis and recurrent tonsillitis (7). Tonsillectomy for these (evidence-based [7]) indications promises an ascertainable change in quality of life, yet data on this topic are sparse; therefore, we decided to carry out this study (8).

The frequency of medication use, the frequency of visits to a doctor, and the number of days off work served as measures of the economic benefit of tonsillectomy. We employed the GBI as well as a questionnaire of our own formulation regarding consumption of analgesics and episodes of pain. These parameters served to estimate the clinical success of the intervention.

The age distribution of our patients, with an average age of 28 years, is comparable with that in other studies in which exclusively adult patients were investigated (4, 15– 18).

The striking female : male ratio of 3.5 : 1 tendentially agrees with reports in the literature (up to 2.5 : 1) (5, 15, 17), but there is no apparent explanation for such a high predominance of women in this study.

Compared with other studies, the 85% response rate among our patients is high. Rates between 26% and 56% have been reported in the literature (5, 15, 16). The highest rate was achieved in the most recent study (16), in which the investigators, like us, not only wrote to the participants but tried, sometimes repeatedly, to reach them by telephone. These authors therefore estimated the selection effect that can be expected for call-back protocols (occupational mobility, migration, death, etc.) as slight in their case.

Tonsillectomy significantly decreased the number of attacks of tonsillitis or sore throat (p = 0.0001). Wolfensberger and Mund (7) also observed a significant reduction from an average of six to two episodes of sore throat per year. Witsell et al.’s weighted analysis (17), though not directly comparable, comes to similar conclusions. The authors ascertained by means of a yes/no question whether their patients experienced more than two episodes of sore throat per month. In summary, however, they too found a statistically significant decrease in symptoms.

The consumption of analgetics and antibiotics also fell significantly after tonsillectomy (p<0.001 and p<0.01 respectively). These findings agree with those of previous studies (4, 5, 16).

The number of visits to the family doctor or ENT specialist decreased significantly over the course of the study period. Mui et al. (5) saw a significant average decrease from four to 0.4 consultations. Bhattacharyya et al. (4, 16) described an even greater reduction in the average number of physician visits, from 5.8 before tonsillectomy to 0.3 thereafter. This difference, too, was significant.

The number of days off work also sank significantly. Bhattacharyya et al. (4) observed a significant reduction from eight days preoperatively to 0.5 days after tonsillectomy. Wolfensberger and Mund (7) described a significant decrease from an average of 7.4 to 1.6 days off work per year.

All studies concur in finding an improvement or reduction in the parameters investigated.

Bhattacharyya and Kepnes (4) used a break-even time analysis to assess the economic benefit of tonsillectomy. They showed that with regard to total costs (medical and work-related), the point when benefits began to outweigh costs (break-even point) came 2.7 years after surgery.

The authors of this study used the GBI to verify the change in patients’ quality of life after tonsillectomy (9). All comparable studies yielded high overall GBI scores and high scores for the “general” subscale. It appears that tonsillectomy improves not only physical health in particular, but also, and especially, the patients’ attitude to life in general or satisfaction with life, as well as contacts and relationships in their social environment (e.g., in the workplace and in public). This can be explained by a strengthening of social ties through fewer episodes of illness. Similar findings were reported by Baumann et al. (18) and Schwentner et al. (15). These authors also observed the highest GBI scores for the subscales “general” and “physical health”, although there was no change in relationships with family and friends. Therefore, it seemed not to be first and foremost a matter of secondary gain. Bhattacharyya et al. (16), in contrast, found only a slight improvement in physical health (+9), compared with the subscales for attitude to life in general (+35) and relationships to family and friends (+14). The authors found no convincing explanation for these differences.

Our study differs from previous investigations by virtue of the high response rate and the additional parameters: numbers of days off work, visits to a doctor, episodes of sore throat, and frequency of medication use.

In common with the authors of three earlier investigations (4, 16– 18), we opted for a retrospective approach, questioning the patients about the quality of life following tonsillectomy by means of the GBI.

This study has some potential limitations. First, we used an additional questionnaire that has not been validated. Second, the observation period was only 14 months—although Schwentner et al. (16) saw no difference in GBI findings with various durations of follow-up. Third, the survey was retrospective. Although such investigations yield very sensitive results (19), they are not always comparable with prospective studies that measure individual parameters at defined time points. Fourth, even with a response rate of 85% it cannot be ruled out that the patients who participated were exclusively those with a positive postoperative course. Fifth, the patients questioned included a low proportion of men, although separate analysis revealed no significant differences in findings between the sexes. Finally, the number of cases may be deemed small.

Our study shows that in adults with at least three attacks of tonsillitis per year, tonsillectomy is likely to distinctly improve quality of life and has discernible economic effects (lower consumption of medication, fewer days off work, and fewer visits to a doctor).

Key Messages.

• In the study presented here, tonsillectomy for recurrent tonsillitis was associated with improved quality of life, lower medication use, fewer days off work than beforehand, a reduced number of visits to a doctor, and lower health-related costs.

• Possible limitations of this study are: lack of validation of one of the questionnaires, the 14-month duration of follow-up, the retrospective approach, the response rate of 85%, the high proportion of women, and the number of cases (n = 114).